Mitigating Challenges in Cloud Anomaly Detection Using an Integrated Deep Neural Network-SVM Classifier Model

Abstract

Cloud workflows remain vulnerable to complex non-linear threats despite existing security solutions. This paper proposes a deep learning model that integrates Support Vector Machines (SVM) algorithm to enhance the security of cloud workflows. The proposed model combines the strengths of SVM’s robust classification capabilities with the flexibility and generalization abilities of deep learning models. The model consists of two main components: a deep neural network (DNN) for feature extraction and an SVM classifier for anomaly detection. The DNN is trained on a large dataset of normal workflow patterns to learn the underlying features that distinguish normal from anomalous behavior. Once the DNN has extracted the relevant features, the SVM classifier is used to classify the workflow patterns as normal or anomalous. The proposed model offers several advantages over traditional anomaly detection methods. The paper also discusses the performance parameters and metrics used to evaluate the effectiveness of proposed deep learning (DL) methods in cloud computing cybersecurity.